Semantic Modelling using Semantic Web Technology
17 likes1,860 views
The document is a detailed overview of the Semantic Web, covering its ideology, key technologies like RDF, RDFS, OWL, and SPARQL, and the importance of ontologies for knowledge representation. It discusses various syntaxes, data integration, and reasoning processes involved in the Semantic Web, emphasizing the formal semantics for dependable reasoning and reusability. Additionally, it outlines resources, tools, and potential applications within the context of the Semantic Web.
Semantic Modelling using Semantic Web Technology
- 1. Semantic Modelling WorkshopSemantic Web TechnologyRinkeHoekstrahoekstra@few.vu.nlSemantic Web Rubik's Cube by dullhunk at flickr under a cc-license. Thanks!
- 3. OverviewSemantic Web: backgroundIdeologySemantic Web and Linked DataKnowledge Representation and OntologyQuick overviewRDF and RDFSOWL 2 DLSKOSSPARQLResourcesWrap UpTutorial information
- 4. Background InformationThe Web as a Knowledge Base
- 5. The Semantic Web IdeologyIdentity is everythingPartial solutions are great too!Layer cake/Web Stack2/22/105OWL
- 6. Key AspectsWorld Wide Web ConsortiumGlobally unique identifiersURI and IRIStraightforward data integrationInterdependent languagesResource Description Framework (RDF)RDF Schema (RDFS)Web Ontology Language (OWL), and OWL 2SPARQL Query Language for RDF (SPARQL)Semantic Web Rule Language (SWRL)Rule Interchange Format (RIF)Simple Knowledge Organization System (SKOS)
- 7. Semantic Web or Linked Data?Initially
- 8. `Metadata’ for web pages
- 9. Since ~2006
- 10. `Web of Data’
- 11. Semantic web as data source in its own right
- 12. Linked Data
- 15. Common schema and identification mechanism
- 16. A single query language2/22/107
- 18. 2/22/109November 2009: 13.1 Billion triples, 142 Million links
- 19. But… not just WebHistorySemantic Networks and Frame Systems (‘70-ies)Formal Knowledge Representation (KL-One)Description Logics (DL)Model Theoretic SemanticsLogic programmingStrongly based in methodologyKnowledge sharing and reuseOntologiesFormal semantics
- 20. Why Ontology?Knowledge bases that mix procedural and declarative knowledge are extremely hard to reuseOntologies capture the `domain theory’ of a KBS: that what a KBS knows aboutProcedures capture what a KBS does with that knowledgeProblem Solving MethodsSlight problem:Rules look like procedures, but can also capture declarative knowledge
- 21. Why Formal Semantics?Standard inferences, dependable reasoningReusable KBS components should guarantee:Soundnessall answers given should be correctCompletenessthe system should give all answersTimelinessthe system should answer within reasonable timeConsistencythe knowledge in a KB should be provably correct
- 22. WARNINGReasoners on the Semantic Web adopt the Open World Assumption as opposed to the Closed World Assumption of many other languagesThis means that:If a reasoner cannot infer that a conclusion holds, it stays silent, and will not report that it does not hold.This is contrary to the negation as failure (NAF) feature of many rule languages.Why?NAF hinders reusability on the Web because conclusions may change depending on new information.
- 23. IngredientsOntology (TBox)Classesconcepts in the domain (≈ DB tables)Propertiesrelations that hold between individuals, or between an individual and a literal value (≈ DB columns)Axioms (only in OWL)restrictions on relations that may hold between instances of a certain class.Assertions (ABox)Individualsinstances of classes in the ontology (≈ DB records)Property assertionsrelations that hold between instances (≈ DB record entries)Rules
- 24. Architecture
- 26. Multiple SyntaxesRDFRDF/XMLVerbose, does not make an exact ‘fit’ with RDF data modelTurtleHuman readableRDFaEmbedded in XHTML as attributesTriXSupports ‘named graphs’OWLAll RDF syntaxes, though RDF/XML is normativeOWL XMLXSLT-ableFunctional SyntaxMapping to Structural Specification (UML MOF)Manchester SyntaxHuman readable, used by most editors (Protégé, TopBraid)
- 27. 2/22/1018URI’sand NamespacesURI: Universal Resource Identifier... Just an identifierhttp://www.leibnizcenter.org/people#joostURL: Universal Resource Locationhttp://www.leibnizcenter.org/people#joostURN: Universal Resource Nameurn:leibnizcenter:people:joostNamespaceA ‘space’ in which all locally defined names are unique, e.g.:http://www.leibnizcenter.org/people#joosthttp://www.hcs.science.uva.nl/staff#joostBasis for `trust’
- 28. RDF SchemaRDF is triple-based<subject,predicate,object>Resourcesrdf:Resourcerdfs:Classrdf:Propertyrdfs:LiteralPropertiesrdf:typerdfs:subClassOfrdfs:subPropertyOfrdfs:domainrdfs:rangerdfs:commentrdfs:labelRDFS Reasoning
- 30. Infer class membership based on domain and rangeRDFS Example@prefix rdf:<http://www.w3.org/1999/02/22-rdf-syntax-ns#>.@prefix rdfs:<http://www.w3.org/TR/rdf-schema/>.@prefixr:<http://www.overheid.nl/regering#>.@prefixk:<http://www.overheid.nl/kabinet#>.@prefix :<http://www.overheid.nl/personen#>.r:Premieraowl:Class;rdfs:subClassOfr:Minister.r:is_premier_vanaowl:ObjectProperty;rdfs:domainr:Premier;rdfs:ranger:Regering.:balkenendear:Premier;r:name ”Jan Peter Balkenende”^^xsd:string;r:is_premier_van k:balkenende_4 ;r:is_premier_van k:balkenende_3 ;r:is_premier_van k:balkenende_2 ;r:is_premier_van k:balkenende_1 ;k:balkenende_4 ar:Regering.2/22/1020
- 32. OWL 2 DL ReasoningConsistency and coherency checkingAre all axioms and assertions consistent?ClassificationDetermine subclass relations between classesInfer disjointness and equivalence of classesRealizationDetermine class membership for individualsInfer new property relations with other individuals
- 33. OWL 2 DL (1)owl:Thing and owl:NothingClass axiomsIntersection, union, disjointness, equivalence:Minister disjointWith:KamerlidsomeValuesFrom (existential):Omnivore subClassOf:Animal and (:eats some :Vegetable)allValuesFrom (universal):Herbivore subClassOf:Animal and (:eats only :Vegetable)hasValue:CowsCalledBettyequivalentTo:Cow and (:name has “Betty”)Self:Narcissist equivalentTo:likes Self
- 34. OWL 2 DL (2)Class axioms (cont’d)(Qualified) Cardinality:Premier equivalentTo:is_premier_vanmin 1 :Kabinet:KabinetsubClassOf:heeft_premiermax 1 :PremierProperty axiomsTransitive, chainsInverseFunctional, inverse functionalReflexive, irreflexiveSymmetric, asymmetricDisjoint, equivalentProperty typesObject Property vs. Datatype Property
- 35. OWL 2 ProfilesSubsets of OWL 2 syntax that have desirable computational propertiesOWL 2 QLOptimized for ontologies with many instancesQuery answeringOWL 2 ELOptimized for ontologies with many classesOWL 2 RLImplementable using rule-based technology
- 36. VocabulariesRDFS and OWL 2 DL semantics often too muchSimple method for describing taxonomiesSimple Knowledge Organization SystemLifting existing KOS’s to the Semantic WebEvery skos:Concept is an OWL individualLightweight semantic relations: broader, narrower, and related.Lightweight mapping relations between skos:ConceptSchemes.JURIX 2009
- 37. Querying: SPARQLRDF Repositories exposed via SPARQL EndpointsQuery typesSELECT … WHERE …CONSTRUCT … WHERE … ( similar to rules!)DELETE … WHERE …ExamplePREFIXr: <http://www.overheid.nl/regering#>PREFIXk: <http://www.overheid.nl/kabinet#>SELECT ?nameWHERE { ?xr:is_premier_van k:balkenende_4. ?xr:name?name }?xbinds with <http://www.overheid.nl/personen#balkenende> ?name binds with“Jan Peter Balkenende”^^xsd:string2/22/1027
- 38. Resources: DocumentationW3C Semantic Web portalhttp://www.w3.org/standards/semanticweb/Start withRDF Primerhttp://www.w3.org/TR/rdf-primer/OWL2 Document Overview & Primerhttp://www.w3.org/TR/owl-overview/http://www.w3.org/TR/owl-primer/SPARQLhttp://www.w3.org/TR/rdf-sparql-query/SKOS Primerhttp://www.w3.org/TR/skos-primer/Rule Interchange Formathttp://www.w3.org/2005/rules/wiki/RIF_Working_Group
- 39. Resources: ToolsEditorsTopBraid Composer (RDF+OWL)Protégé 4 (OWL Only)ReasonersPellet, Fact++, OWLIM, Racer, HermiT, CELRDF RepositoriesOpenRDF Sesame, 4Store/5Store, Virtuoso, MulgaraRDB to RDF bridgesD2RQ
- 40. Wrap Up: BenefitsSingle `database schema’ eases query writingGlobal identifiers facilitate data integrationLayered approach eases extensibilityFormal semantics ensure dependability
- 41. Wrap Up: Tutorial Information2 Hour sessionTopBraid Composer Standard (trial license)What we’ll do:Understanding an ontology & using a reasonerData integration using SPARQLSimple OWL class definitionsAdvanced OWL class definitions (extra)